This paper is a theoretical analysis of eidetic imagery, based upon the author's ten-year study of elementary-school-aged children. The presence of eidetic imagery is inferred from reports of persisting visual images of stimuli when they are no longer in view. According to the criteria for differentiating eidetic images from afterimages, eidetic images should occur even when saccadic eye movements are made during exposure to the stimulus; it should be possible to make saccadic eye movements while one is reporting the image without the image also moving; the image should last long, and it should be positive. The criteria for differentiating eidetic images from nonvisual memorial representations include: reports of seeing an image projected onto a surface in space, the consistent use of present tense when reporting images as opposed to past tense when reporting from nonvisual memory, and the ability to superimpose two images and report the composite image.

Eidetic images are only available to a small percentage of children 6–12 years old, and are virtually nonexistent in adults. However, extensive research has failed to demonstrate consistent correlates between the presence of eidetic imagery and any cognitive, intellectual, neurological, or emotional measure. The negative correlation between eidetic imagery and age has prompted hypotheses to explain eidetic imagery as a developmentally less mature memorial representation, which is gradually replaced by more abstract representations as the child acquires abstract thought, reading, and more advanced cognitive abilities. The evidence in the present review casts doubt on this hypothesis on numerous grounds: an extensive longitudinal study over the entire span of elementary school years found that eidetic abilities remain remarkably stable; there is no correlation between eidetic imagery and abstract thinking or reading performance; there is no higher incidence in preschool ages, among retarded or brain-injured subjects, or among illiterate subjects in crosscultural studies.

It is concluded that we should not abandon work on eidetic imagery or simply force it into a preconceived mold of what memory must be, but rather, expand work on the phenomenological indicators of perception and memory.